The demand for bandwidth continues to grow exponentially on fiber-optic superhighways due to applications such as data communications and the Internet. Consequently, there is much effort at exploiting the bandwidth of optical fibers by using higher speeds per channel—so-called time-division multiplexed systems—and multiple wavelengths of light—so-called wavelength-division multiplexing (WDM).
Most of the fiber-optic networks currently deployed use one of two kinds of fiber: standard single-mode fiber (“standard” fiber) or dispersion-shifted fiber (DSF). Standard fiber has a zero dispersion wavelength around 1310 nm, and the dispersion is primarily resulting from the inherent glass dispersion. Most of the terrestrial network in the US and most of the world is, in fact, based on standard fiber. In DSF, on the other hand, waveguide dispersion is used to shift the zero dispersion wavelength to longer wavelengths. A conventional DSF will have a zero dispersion wavelength at 1550 nm, coinciding with the minimum loss in a fused silica fiber. However, the zero dispersion wavelength can be shifted around by varying the amount of waveguide dispersion added. DSF is used exclusively in two countries, Japan and Italy, as well as in new long-haul links.
The limiting factors for a fiber-optic transmission line include loss, dispersion and gain equalization. Loss refers to the fact that the signal attenuates as it travels in a fiber due to intrinsic scattering, absorption and other extrinsic effects such as defects. Optical amplifiers, for example, can be used to compensate for the loss. Dispersion means that different frequencies of light travel at different speeds, and it comes from both the material properties and waveguiding effects. When using multi-wavelength systems and due to the nonuniformity of the gain with frequency, gain equalization is required to even out the gain over the different wavelength channels.
The typical solution to overcoming these limitations is to place periodically in a transmission system elements to compensate for each of these problems. For example, FIG. 1 shows that a dispersion compensator 40 can be used to cancel the dispersion, an optical amplifier 50 can be used to balance the loss, and a gain equalization element 60 can be used to flatten the gain. Examples of dispersion compensators include chirped fiber gratings and dispersion compensating fiber (DCF). Examples of optical amplifiers include erbium-doped fiber amplifiers (EDFAs), Raman amplifiers, and non-linear fiber amplifiers (NLFAs). U.S. Pat. No. 5,778,014 discloses Sagnac Raman amplifiers and cascade lasers.
Finally, examples of gain equalizers include Mach-Zehnder interferometers and long period gratings. Rather than building a system out of these individual components, it may be easier and more cost effective to combine two or more of the functions in FIG. 1 into a single component as shown in U.S. Pat. No. 5,887,093.
Another problem that arises in WDM systems is interaction or cross-talk between channels through nonlinearities in the fiber. In particular, four-wave mixing (4WM) causes exchange of energy between different wavelength channels, but 4WM only phase matches near the zero dispersion wavelength. Consequently, if a fiber link is made from conventional DSF, it is difficult to operate a WDM system from around 1540-1560 nm. This turns out to be quite unfortunate because typical EDFA's have gain from 1535-1565 nm, and the more uniform gain band is near 1540-1560 nm. A second fiber nonlinearity that can be troublesome is modulation instability (MI), which is 4WM where the fiber's nonlinear index-of-refraction helps to phase match. However, MI only phase matches when the dispersion is positive or in the so-called soliton regime. Therefore, MI can be avoided by operating at wavelengths shorter than the zero dispersion wavelength.
As the bandwidth utilization over individual fibers increases, the number of bands used for transmission increases. For WDM systems using a number of bands, additional complexities arise due to interaction between and amplification in multi-band scenarios. In particular, particular system designs are needed for Raman amplification in multi-band transmission systems. First, a new nonlinearity penalty arises from the gain tilt from the Raman effect between channels. This arises because long wavelength channels tend to rob energy from the short wavelength channels. Therefore, a means of minimizing the gain tilt on existing channels with the addition of new WDM channels is required.
To minimize both the effects of 4WM and Raman gain tilt, another technical strategy is to use distributed Raman amplification. In a WDM system with multi-bands, a complexity arises from interaction between the different pumps along the transmission line.
There is a need for broadband amplifiers that span more than 40 nm of bandwidth. There is a further need for broadband amplifiers that amplify wavelengths in the low loss window of optical fibers. There is yet a further need for broadband amplifiers that reduce the cost per wavelength. There is a need for broadband amplifiers without band splitters and combiners and guard bands between the bands.